Cheap CFL teardown after failure

[tom66]

Tesco 11W CFL, cost around 50p when bought (when gov't was subsidising them a few years ago... now they are about £4 each.)

This CFL started making a buzzing sound whilst in use and to my ears this was very annoying and sounded like a failing CFL. Replaced it with a Philips bulb, but I'm not confident that will last long either from what I've seen.

It is rated for 10 years but it has serious discolouration around the bulb base so my guess is it had a couple of months left (it is about 2 years and is used 8-10 hours per day.) So, 10 years is a complete lie, no way this bulb could make it.

Some interesting points:
- No fuse. Not even a fusible resistor! The closest to a fuse is the input inductor, which is the only EMI filtering this bulb has; that might have a fusing current, but I'm skeptical it would meet CE.
- No thermal fuse, so failure mode will be fun with flames.
- X cap is a poly cap, not a genuine flame-proof X cap, and it is before the inductor, so there is no fusing to protect it. More firey fun!
- It uses a 2.8µF cap. No, that's not a typo - not 2.7µF. The manufacturer obviously was too cheap to use 3.3µF. At least they use a 400V cap - I've seen a few bulbs using 350V caps, which sounds "okay" until you work out that the maximum voltage in the UK allowed is 253V which gives 357V rectified, which is exceeding the cap ratings. I've seen 251V from the socket here. Cap is an Aishu, I've seen this brand in some cheap TVs.
- Royer oscillator transistors are TO-92 13002's. They should be at the very least TO-253 13005's, or even better TO-220 13007's.
- Transformer is discoloured from heat.
- The coating of the bulb has worn out on one specific place, I'm not sure why.
- A lot of the components were pushed over when I first opened it up.
- One of the green caps has discoloured considerably.

No idea how this passed CE - if it did at all! Maybe CE was faked.

I don't like these bulbs because they toast their ballasts and they do not last the 10 years claimed.

[tom66]

Checked the cap - 2.69µF 3.9 ohms. I think the ESR is too high for comfort, probably the cause of the problem. No series on the Aishi cap, so no idea on what the ESR should be.

[kc8adu]

the fuse is the .5 ohm resistor.
and the buzzing is likely the cap.
my neighbors huge cfl plant bulbs buzzed just before the caps gave up.

[ben7]

Originally Posted by kc8adu

the fuse is the .5 ohm resistor.
and the buzzing is likely the cap.
my neighbors huge cfl plant bulbs buzzed just before the caps gave up.

NEVERMIND THIS! {Hmm, I think there is no fuse, those resistors are likely the emitter resistors for the transistors. If the rectifier diodes short, then there will be sparks and flames, but if a transistor shorts, there will be a small pop and thats it.}
I just looked closely and there is a resistor 'fuse' in series with the power wires connecting to the screw base.

BTW check for cracked solder joints on the inductor, and replace the 2.8uF cap with a bigger one. I noticed in cfls that if the inductor has bad solder joint(s), sometimes there will be little arcs, so it buzzes.

-Ben

[Th3_uN1Qu3]

You will never see 13007s in a CFL. Ever. Nowadays they get by with 13001s... I have seen some with TO-126 transistors but that was in older bulbs.

The ESR on that cap is so-so. Not exceptional but okay. There isn't too much it needs to filter anyway. The likely reason for the buzzing noise is the green poly cap in series with the bulb. That is, unless the buzzing is at mains frequency or its harmonics, in which case it's the primary cap. If it's more of a whine then the green cap is to blame. Those green caps are a common failure point in CFL ballasts as they're almost always underrated. As soon as the bulb starts pulling a bit more current due to age, that capacitor quickly overheats and fails.

[tom66]

Originally Posted by Th3_uN1Qu3

You will never see 13007s in a CFL. Ever. Nowadays they get by with 13001s... I have seen some with TO-126 transistors but that was in older bulbs.

The ESR on that cap is so-so. Not exceptional but okay. There isn't too much it needs to filter anyway. The likely reason for the buzzing noise is the green poly cap in series with the bulb. That is, unless the buzzing is at mains frequency or its harmonics, in which case it's the primary cap. If it's more of a whine then the green cap is to blame. Those green caps are a common failure point in CFL ballasts as they're almost always underrated. As soon as the bulb starts pulling a bit more current due to age, that capacitor quickly overheats and fails.

I got confused between TO-253 and TO-126; but there are a few bulbs with TO-126:

http://www.pavouk.org/hw/lamp/en_index.html

I swear I've seen older ones with TO-220 but I'll have to check.

I just checked and there is a fusible resistor inline with the AC input, but it's not on the PCB. Still, there is no thermal fuse like manufacturers claim their bulbs have.

[selldoor]

They dont need a fuse - they rely on the one in your head that blows with the noise or flicker long before the bulb gives up.

[ben7]

LOL, did a job with an electrician the other day installing cfls, and they say on the box they will last 25,000 hours - NO WAY JOSE! In fact one of them had trouble starting up, it took a few seconds, and one filament burned bright orange; I bet we will be replacing that one soon! xD

-Ben

[lti]

Originally Posted by tom66

It is rated for 10 years but it has serious discolouration around the bulb base so my guess is it had a couple of months left (it is about 2 years and is used 8-10 hours per day.) So, 10 years is a complete lie, no way this bulb could make it.

Older CFLs (the first ones made) really could last that long. Since then, the quality has dropped so much that it is actually normal for a bulb to fail catastrophically after only a few months.

Surprisingly, I still have one of the "4 for $10" Greenlite CFLs that still works. When it dies, I might take it apart. The other three died quickly. One failed "normally," one failed in a safe manner (it just stopped lighting up without any external failure signs like discolored plastic or a burning smell), and one started flickering slowly and stopped working a few seconds later.

[budm]

The tube itself will last, it is the electronics that failed, I just hate it when the new LED bulbs are being claimed to last 30~50,000 Hours, sure!, the electrolytic caps will really last that long in small enclosure? the weakest link will always take out the whole unit.

[Evil Lurker]

I have one CFL that managed to last almost 3 years in my living room ceiling fan... pretty much all the rest have gone to electronics heaven. Out of the three siblings that died, I did an autopsy on two of them and was horrified at what I found... the insulation on the lead wires going from the base to the PCB was almost burned off. This did not look like your typical quick shorting out and the wire got too hot but rather prolonged heat exposure decomposition.

I think this is the biggest cause of failure for CFL's... they literally cook themselves to death. Any cap that can withstand a year of 8 hours a day upside down in a recessed fixture is A-OK in my book.

I wonder how much of a safety issue it would be to simply cut a few slots around the base of the bulb to let them breathe... I bet that could potentially double their useful lifespan.

[PeteS in CA]

Couple interesting things I noted in the pix. The solder joints for the transformer wires onto the PCB terminals look crappy (if not toasty badness), assuming there are solder joints and not just wire-wrap style contacts. The PCB silkscreen shape for the inverter transistor looks like a TO-126 or TO-220.

[PeteS in CA]

Many moons ago in my first incarnation here I posted several tech news articles about CFL quality concerns. One particular problem was that they had poor heat dissipation in some common mounting angles. I'll dig around, see if I can find the threads.

[lti]

I have a Commercial Electric CFL that was purchased when CFLs first came out. It still works. It came in a package of two bulbs, and the second bulb failed last year. They were powered on for an average amount of time. The bulb that failed just stopped lighting up without flickering, smoking, making noises, or discoloring the base. It still looked like a new bulb.

[PeteS in CA]

Did some digging around, didn't find anything more that I posted quite a bit (Duh-uhhh!).

[PeteS in CA]

Well, I nosed around in another discussion board and came up with:

http://www.edn.com/blog/PowerSource/...cent_light.php

http://www.edn.com/blog/PowerSource/...ed_by_CFLs.php

http://www.nytimes.com/2009/03/28/bu...s.html?_r=3&hp

http://www.edn.com/blog/PowerSource/...wer_factor.php

[lti]

That New York Times article said that CFLs started to be subsidized by energy companies in 1998. I didn't know that CFLs had been manufactured for that long. I only started seeing them around 2003 when the Commercial Electric CFLs I mentioned above were purchased.

[tom66]

Why would an energy company subsidise something that causes a reduction in energy usage?

[larrymoencurly]

Originally Posted by tom66

Why would an energy company subsidise something that causes a reduction in energy usage?

It can be cheaper than having to build or buy more generation capacity.

We have some ancient Philips CFLs where the ballast and bulb are separate, and they flicker badly about 3 times when turned on. I think the ballast works at 60 Hz because it's heavy.

[smason]

I wish I had recorded the date I installed mine on, but I have a houseful of Philips Marathon CFLs and they're going strong after at least 5 years now.

ok, one died but it was mounted upside-down and likely died a heat-related death.